Annotation of rpl/lapack/lapack/zgehrd.f, revision 1.17

1.9       bertrand    1: *> \brief \b ZGEHRD
                      2: *
                      3: *  =========== DOCUMENTATION ===========
                      4: *
1.16      bertrand    5: * Online html documentation available at
                      6: *            http://www.netlib.org/lapack/explore-html/
1.9       bertrand    7: *
                      8: *> \htmlonly
1.16      bertrand    9: *> Download ZGEHRD + dependencies
                     10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/zgehrd.f">
                     11: *> [TGZ]</a>
                     12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/zgehrd.f">
                     13: *> [ZIP]</a>
                     14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/zgehrd.f">
1.9       bertrand   15: *> [TXT]</a>
1.16      bertrand   16: *> \endhtmlonly
1.9       bertrand   17: *
                     18: *  Definition:
                     19: *  ===========
                     20: *
                     21: *       SUBROUTINE ZGEHRD( N, ILO, IHI, A, LDA, TAU, WORK, LWORK, INFO )
1.16      bertrand   22: *
1.9       bertrand   23: *       .. Scalar Arguments ..
                     24: *       INTEGER            IHI, ILO, INFO, LDA, LWORK, N
                     25: *       ..
                     26: *       .. Array Arguments ..
                     27: *       COMPLEX*16        A( LDA, * ), TAU( * ), WORK( * )
                     28: *       ..
1.16      bertrand   29: *
1.9       bertrand   30: *
                     31: *> \par Purpose:
                     32: *  =============
                     33: *>
                     34: *> \verbatim
                     35: *>
                     36: *> ZGEHRD reduces a complex general matrix A to upper Hessenberg form H by
                     37: *> an unitary similarity transformation:  Q**H * A * Q = H .
                     38: *> \endverbatim
                     39: *
                     40: *  Arguments:
                     41: *  ==========
                     42: *
                     43: *> \param[in] N
                     44: *> \verbatim
                     45: *>          N is INTEGER
                     46: *>          The order of the matrix A.  N >= 0.
                     47: *> \endverbatim
                     48: *>
                     49: *> \param[in] ILO
                     50: *> \verbatim
                     51: *>          ILO is INTEGER
                     52: *> \endverbatim
                     53: *>
                     54: *> \param[in] IHI
                     55: *> \verbatim
                     56: *>          IHI is INTEGER
                     57: *>
                     58: *>          It is assumed that A is already upper triangular in rows
                     59: *>          and columns 1:ILO-1 and IHI+1:N. ILO and IHI are normally
                     60: *>          set by a previous call to ZGEBAL; otherwise they should be
                     61: *>          set to 1 and N respectively. See Further Details.
                     62: *>          1 <= ILO <= IHI <= N, if N > 0; ILO=1 and IHI=0, if N=0.
                     63: *> \endverbatim
                     64: *>
                     65: *> \param[in,out] A
                     66: *> \verbatim
                     67: *>          A is COMPLEX*16 array, dimension (LDA,N)
                     68: *>          On entry, the N-by-N general matrix to be reduced.
                     69: *>          On exit, the upper triangle and the first subdiagonal of A
                     70: *>          are overwritten with the upper Hessenberg matrix H, and the
                     71: *>          elements below the first subdiagonal, with the array TAU,
                     72: *>          represent the unitary matrix Q as a product of elementary
                     73: *>          reflectors. See Further Details.
                     74: *> \endverbatim
                     75: *>
                     76: *> \param[in] LDA
                     77: *> \verbatim
                     78: *>          LDA is INTEGER
                     79: *>          The leading dimension of the array A.  LDA >= max(1,N).
                     80: *> \endverbatim
                     81: *>
                     82: *> \param[out] TAU
                     83: *> \verbatim
                     84: *>          TAU is COMPLEX*16 array, dimension (N-1)
                     85: *>          The scalar factors of the elementary reflectors (see Further
                     86: *>          Details). Elements 1:ILO-1 and IHI:N-1 of TAU are set to
                     87: *>          zero.
                     88: *> \endverbatim
                     89: *>
                     90: *> \param[out] WORK
                     91: *> \verbatim
                     92: *>          WORK is COMPLEX*16 array, dimension (LWORK)
                     93: *>          On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
                     94: *> \endverbatim
                     95: *>
                     96: *> \param[in] LWORK
                     97: *> \verbatim
                     98: *>          LWORK is INTEGER
                     99: *>          The length of the array WORK.  LWORK >= max(1,N).
1.14      bertrand  100: *>          For good performance, LWORK should generally be larger.
1.9       bertrand  101: *>
                    102: *>          If LWORK = -1, then a workspace query is assumed; the routine
                    103: *>          only calculates the optimal size of the WORK array, returns
                    104: *>          this value as the first entry of the WORK array, and no error
                    105: *>          message related to LWORK is issued by XERBLA.
                    106: *> \endverbatim
                    107: *>
                    108: *> \param[out] INFO
                    109: *> \verbatim
                    110: *>          INFO is INTEGER
                    111: *>          = 0:  successful exit
                    112: *>          < 0:  if INFO = -i, the i-th argument had an illegal value.
                    113: *> \endverbatim
                    114: *
                    115: *  Authors:
                    116: *  ========
                    117: *
1.16      bertrand  118: *> \author Univ. of Tennessee
                    119: *> \author Univ. of California Berkeley
                    120: *> \author Univ. of Colorado Denver
                    121: *> \author NAG Ltd.
1.9       bertrand  122: *
1.16      bertrand  123: *> \date December 2016
1.9       bertrand  124: *
                    125: *> \ingroup complex16GEcomputational
                    126: *
                    127: *> \par Further Details:
                    128: *  =====================
                    129: *>
                    130: *> \verbatim
                    131: *>
                    132: *>  The matrix Q is represented as a product of (ihi-ilo) elementary
                    133: *>  reflectors
                    134: *>
                    135: *>     Q = H(ilo) H(ilo+1) . . . H(ihi-1).
                    136: *>
                    137: *>  Each H(i) has the form
                    138: *>
                    139: *>     H(i) = I - tau * v * v**H
                    140: *>
                    141: *>  where tau is a complex scalar, and v is a complex vector with
                    142: *>  v(1:i) = 0, v(i+1) = 1 and v(ihi+1:n) = 0; v(i+2:ihi) is stored on
                    143: *>  exit in A(i+2:ihi,i), and tau in TAU(i).
                    144: *>
                    145: *>  The contents of A are illustrated by the following example, with
                    146: *>  n = 7, ilo = 2 and ihi = 6:
                    147: *>
                    148: *>  on entry,                        on exit,
                    149: *>
                    150: *>  ( a   a   a   a   a   a   a )    (  a   a   h   h   h   h   a )
                    151: *>  (     a   a   a   a   a   a )    (      a   h   h   h   h   a )
                    152: *>  (     a   a   a   a   a   a )    (      h   h   h   h   h   h )
                    153: *>  (     a   a   a   a   a   a )    (      v2  h   h   h   h   h )
                    154: *>  (     a   a   a   a   a   a )    (      v2  v3  h   h   h   h )
                    155: *>  (     a   a   a   a   a   a )    (      v2  v3  v4  h   h   h )
                    156: *>  (                         a )    (                          a )
                    157: *>
                    158: *>  where a denotes an element of the original matrix A, h denotes a
                    159: *>  modified element of the upper Hessenberg matrix H, and vi denotes an
                    160: *>  element of the vector defining H(i).
                    161: *>
                    162: *>  This file is a slight modification of LAPACK-3.0's DGEHRD
                    163: *>  subroutine incorporating improvements proposed by Quintana-Orti and
                    164: *>  Van de Geijn (2006). (See DLAHR2.)
                    165: *> \endverbatim
                    166: *>
                    167: *  =====================================================================
1.1       bertrand  168:       SUBROUTINE ZGEHRD( N, ILO, IHI, A, LDA, TAU, WORK, LWORK, INFO )
                    169: *
1.16      bertrand  170: *  -- LAPACK computational routine (version 3.7.0) --
1.1       bertrand  171: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
                    172: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
1.16      bertrand  173: *     December 2016
1.1       bertrand  174: *
                    175: *     .. Scalar Arguments ..
                    176:       INTEGER            IHI, ILO, INFO, LDA, LWORK, N
                    177: *     ..
                    178: *     .. Array Arguments ..
                    179:       COMPLEX*16        A( LDA, * ), TAU( * ), WORK( * )
                    180: *     ..
                    181: *
                    182: *  =====================================================================
                    183: *
                    184: *     .. Parameters ..
1.14      bertrand  185:       INTEGER            NBMAX, LDT, TSIZE
                    186:       PARAMETER          ( NBMAX = 64, LDT = NBMAX+1,
                    187:      $                     TSIZE = LDT*NBMAX )
1.1       bertrand  188:       COMPLEX*16        ZERO, ONE
1.16      bertrand  189:       PARAMETER          ( ZERO = ( 0.0D+0, 0.0D+0 ),
1.1       bertrand  190:      $                     ONE = ( 1.0D+0, 0.0D+0 ) )
                    191: *     ..
                    192: *     .. Local Scalars ..
                    193:       LOGICAL            LQUERY
1.14      bertrand  194:       INTEGER            I, IB, IINFO, IWT, J, LDWORK, LWKOPT, NB,
1.1       bertrand  195:      $                   NBMIN, NH, NX
                    196:       COMPLEX*16        EI
                    197: *     ..
                    198: *     .. External Subroutines ..
                    199:       EXTERNAL           ZAXPY, ZGEHD2, ZGEMM, ZLAHR2, ZLARFB, ZTRMM,
                    200:      $                   XERBLA
                    201: *     ..
                    202: *     .. Intrinsic Functions ..
                    203:       INTRINSIC          MAX, MIN
                    204: *     ..
                    205: *     .. External Functions ..
                    206:       INTEGER            ILAENV
                    207:       EXTERNAL           ILAENV
                    208: *     ..
                    209: *     .. Executable Statements ..
                    210: *
                    211: *     Test the input parameters
                    212: *
                    213:       INFO = 0
                    214:       LQUERY = ( LWORK.EQ.-1 )
                    215:       IF( N.LT.0 ) THEN
                    216:          INFO = -1
                    217:       ELSE IF( ILO.LT.1 .OR. ILO.GT.MAX( 1, N ) ) THEN
                    218:          INFO = -2
                    219:       ELSE IF( IHI.LT.MIN( ILO, N ) .OR. IHI.GT.N ) THEN
                    220:          INFO = -3
                    221:       ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
                    222:          INFO = -5
                    223:       ELSE IF( LWORK.LT.MAX( 1, N ) .AND. .NOT.LQUERY ) THEN
                    224:          INFO = -8
                    225:       END IF
1.14      bertrand  226: *
                    227:       IF( INFO.EQ.0 ) THEN
                    228: *
                    229: *        Compute the workspace requirements
                    230: *
                    231:          NB = MIN( NBMAX, ILAENV( 1, 'ZGEHRD', ' ', N, ILO, IHI, -1 ) )
                    232:          LWKOPT = N*NB + TSIZE
                    233:          WORK( 1 ) = LWKOPT
                    234:       ENDIF
                    235: *
1.1       bertrand  236:       IF( INFO.NE.0 ) THEN
                    237:          CALL XERBLA( 'ZGEHRD', -INFO )
                    238:          RETURN
                    239:       ELSE IF( LQUERY ) THEN
                    240:          RETURN
                    241:       END IF
                    242: *
                    243: *     Set elements 1:ILO-1 and IHI:N-1 of TAU to zero
                    244: *
                    245:       DO 10 I = 1, ILO - 1
                    246:          TAU( I ) = ZERO
                    247:    10 CONTINUE
                    248:       DO 20 I = MAX( 1, IHI ), N - 1
                    249:          TAU( I ) = ZERO
                    250:    20 CONTINUE
                    251: *
                    252: *     Quick return if possible
                    253: *
                    254:       NH = IHI - ILO + 1
                    255:       IF( NH.LE.1 ) THEN
                    256:          WORK( 1 ) = 1
                    257:          RETURN
                    258:       END IF
                    259: *
                    260: *     Determine the block size
                    261: *
                    262:       NB = MIN( NBMAX, ILAENV( 1, 'ZGEHRD', ' ', N, ILO, IHI, -1 ) )
                    263:       NBMIN = 2
                    264:       IF( NB.GT.1 .AND. NB.LT.NH ) THEN
                    265: *
                    266: *        Determine when to cross over from blocked to unblocked code
                    267: *        (last block is always handled by unblocked code)
                    268: *
                    269:          NX = MAX( NB, ILAENV( 3, 'ZGEHRD', ' ', N, ILO, IHI, -1 ) )
                    270:          IF( NX.LT.NH ) THEN
                    271: *
                    272: *           Determine if workspace is large enough for blocked code
                    273: *
1.14      bertrand  274:             IF( LWORK.LT.N*NB+TSIZE ) THEN
1.1       bertrand  275: *
                    276: *              Not enough workspace to use optimal NB:  determine the
                    277: *              minimum value of NB, and reduce NB or force use of
                    278: *              unblocked code
                    279: *
                    280:                NBMIN = MAX( 2, ILAENV( 2, 'ZGEHRD', ' ', N, ILO, IHI,
                    281:      $                 -1 ) )
1.14      bertrand  282:                IF( LWORK.GE.(N*NBMIN + TSIZE) ) THEN
                    283:                   NB = (LWORK-TSIZE) / N
1.1       bertrand  284:                ELSE
                    285:                   NB = 1
                    286:                END IF
                    287:             END IF
                    288:          END IF
                    289:       END IF
                    290:       LDWORK = N
                    291: *
                    292:       IF( NB.LT.NBMIN .OR. NB.GE.NH ) THEN
                    293: *
                    294: *        Use unblocked code below
                    295: *
                    296:          I = ILO
                    297: *
                    298:       ELSE
                    299: *
                    300: *        Use blocked code
                    301: *
1.14      bertrand  302:          IWT = 1 + N*NB
1.1       bertrand  303:          DO 40 I = ILO, IHI - 1 - NX, NB
                    304:             IB = MIN( NB, IHI-I )
                    305: *
                    306: *           Reduce columns i:i+ib-1 to Hessenberg form, returning the
1.8       bertrand  307: *           matrices V and T of the block reflector H = I - V*T*V**H
1.1       bertrand  308: *           which performs the reduction, and also the matrix Y = A*V*T
                    309: *
1.14      bertrand  310:             CALL ZLAHR2( IHI, I, IB, A( 1, I ), LDA, TAU( I ),
                    311:      $                   WORK( IWT ), LDT, WORK, LDWORK )
1.1       bertrand  312: *
                    313: *           Apply the block reflector H to A(1:ihi,i+ib:ihi) from the
1.8       bertrand  314: *           right, computing  A := A - Y * V**H. V(i+ib,ib-1) must be set
1.1       bertrand  315: *           to 1
                    316: *
                    317:             EI = A( I+IB, I+IB-1 )
                    318:             A( I+IB, I+IB-1 ) = ONE
1.16      bertrand  319:             CALL ZGEMM( 'No transpose', 'Conjugate transpose',
1.1       bertrand  320:      $                  IHI, IHI-I-IB+1,
                    321:      $                  IB, -ONE, WORK, LDWORK, A( I+IB, I ), LDA, ONE,
                    322:      $                  A( 1, I+IB ), LDA )
                    323:             A( I+IB, I+IB-1 ) = EI
                    324: *
                    325: *           Apply the block reflector H to A(1:i,i+1:i+ib-1) from the
                    326: *           right
                    327: *
                    328:             CALL ZTRMM( 'Right', 'Lower', 'Conjugate transpose',
                    329:      $                  'Unit', I, IB-1,
                    330:      $                  ONE, A( I+1, I ), LDA, WORK, LDWORK )
                    331:             DO 30 J = 0, IB-2
                    332:                CALL ZAXPY( I, -ONE, WORK( LDWORK*J+1 ), 1,
                    333:      $                     A( 1, I+J+1 ), 1 )
                    334:    30       CONTINUE
                    335: *
                    336: *           Apply the block reflector H to A(i+1:ihi,i+ib:n) from the
                    337: *           left
                    338: *
                    339:             CALL ZLARFB( 'Left', 'Conjugate transpose', 'Forward',
                    340:      $                   'Columnwise',
1.14      bertrand  341:      $                   IHI-I, N-I-IB+1, IB, A( I+1, I ), LDA,
                    342:      $                   WORK( IWT ), LDT, A( I+1, I+IB ), LDA,
                    343:      $                   WORK, LDWORK )
1.1       bertrand  344:    40    CONTINUE
                    345:       END IF
                    346: *
                    347: *     Use unblocked code to reduce the rest of the matrix
                    348: *
                    349:       CALL ZGEHD2( N, I, IHI, A, LDA, TAU, WORK, IINFO )
1.14      bertrand  350:       WORK( 1 ) = LWKOPT
1.1       bertrand  351: *
                    352:       RETURN
                    353: *
                    354: *     End of ZGEHRD
                    355: *
                    356:       END

CVSweb interface <joel.bertrand@systella.fr>